Insights
Kimi K3 and DeepSeek V4 expose widening divide over native multimodality
Chinese AI model developers agree on multimodality’s long-term value, but not on when its benefits justify the cost.
10 mins read
Cheng Zi
Cheng Zi is based in China, covering technology- and business-related events. She works with her colleagues at 36Kr Global to provide coverage in both English and Chinese.
The former Tencent and Anuttacon researcher brings experience in human-centric, multimodal video systems.
The AI company said its model surpassed GLM-5.2 by training four billion additional parameters through specialized LoRA adapters.
The fashion retailer’s LATR system has helped suppliers adjust more quickly to shifts in consumer trends and demand.
After building Cece into a profitable app, Cylingo is applying its tech to a home robot designed to read the mood of a household.
As AI enables tiny teams to produce films on modest budgets, China’s established studios are being forced to confront who gets to make movies.
Geely has paused similar projects at other group brands as it directs resources toward Volvo.
Former OpenAI researcher Tian Yonglong will succeed multimodal chief Han Hu, who has reportedly resigned.
The hardware executive succeeds founder Henry Zhou as Pico shifts focus toward mixed reality and spatial computing.